The daily firehose — everything the toolchain shipped today, already filtered.
// HOW THIS ISSUE IS MADE
We read every release from the 174 tools on our watchlist at the source — GitHub and GitLab release notes, vendor release pages and changelogs, project blogs and feeds, vendor press releases, and the source code behind the tag. Bug-fix-only releases and non-product newsroom noise are dropped; what's left is summarized down to the new capability, how to try it, and any screenshots or videos the release itself published. Every entry links to the sources it was built from.
Agno v2.3.12 adds token-count-based context compression support across providers.
└──▷ GET THIS VERSION
$ git clone --branch v2.3.12 https://github.com/agno-agi/agno.git
# already have the repo? check out this version:$ git checkout v2.3.12
›Adds token count based compression via the Compression Manager, enabling context window management across providers based on token limits.
›Content hashing for knowledge ingestion now incorporates name and description fields for URLs, paths, and file_data, allowing multiple distinct content items from the same source to produce unique hashes.
└──▷ BREAKING ON UPGRADE
!Existing knowledge content previously added with name or description fields will have different content hashes under v2.3.12, which may alter skip_if_exists and upsert behavior for that content.
langchain-groq 1.1.1 lets kwargs in with_structured_output override tool_choice and filters unsupported parameters in bind_tools.
└──▷ GET THIS VERSION
$ git clone --branch langchain-groq==1.1.1 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:$ git checkout langchain-groq==1.1.1
›Allows keyword arguments passed to with_structured_output to override the default tool_choice setting, giving callers per-invocation control over tool selection.
›Filters unsupported parameters in bind_tools for Groq, preventing invalid arguments from being forwarded to the API.
langchain-tests 1.1.0 adds invocation model override and stricter usage_metadata chunk validation for standard tests.
└──▷ GET THIS VERSION
$ git clone --branch langchain-tests==1.1.0 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:$ git checkout langchain-tests==1.1.0
›Adds invocation model override capability to standard tests, allowing test configurations to specify a different model for invocation.
›Adds a standard test that ensures only one chunk sets model_name in usage_metadata during streaming responses.
└──▷ BREAKING ON UPGRADE
!The deprecated has_tool_choice property has been removed; any test suite referencing it will break on upgrade.
langchain-anthropic 1.3.0 adds MCP toolset binding, tool search, effort control, computer-use headers, and TypedDict support for built-in tools.
└──▷ GET THIS VERSION
$ git clone --branch langchain-anthropic==1.3.0 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:$ git checkout langchain-anthropic==1.3.0
└──▷ USE IT
Bind an MCP toolset to an Anthropic chat model so the model can call MCP-hosted tools without manually setting beta headers.
python
from langchain_anthropic import ChatAnthropic
llm = ChatAnthropic(model="claude-opus-4-5")
llm_with_tools = llm.bind_tools([mcp_toolset])
Pass provider-specific parameters through a tool using the new extras field on BaseTool.
langchain-core 1.2.0 adds an extras field to BaseTool for attaching arbitrary metadata.
└──▷ GET THIS VERSION
$ git clone --branch langchain-core==1.2.0 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:$ git checkout langchain-core==1.2.0
›Adds extras field to BaseTool, enabling arbitrary key-value metadata to be attached to any tool definition.
langchain-chroma 1.1.0 adds a Search API to the Chroma vector store integration.
└──▷ GET THIS VERSION
$ git clone --branch langchain-chroma==1.1.0 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:$ git checkout langchain-chroma==1.1.0
›Adds a Search API to the Chroma vector store integration, enabling direct search calls through the LangChain Chroma wrapper.
PydanticAI v1.31.0 adds prompt caching for AWS Bedrock, Agent.output_json_schema(), custom MCP clientInfo, and GPT-5.2 support.
└──▷ GET THIS VERSION
$ git clone --branch v1.31.0 https://github.com/pydantic/pydantic-ai.git
# already have the repo? check out this version:$ git checkout v1.31.0
└──▷ USE IT
Retrieve the JSON schema for an agent's structured output to validate or document the expected response shape.
python
from pydantic_ai import Agent
from pydantic import BaseModel
class Answer(BaseModel):
summary: str
confidence: float
agent = Agent('openai:gpt-4o', output_type=Answer)
print(agent.output_json_schema())
Use a plain model name string in LLMJudge instead of a model object, for quick evaluation scripting.
python
from pydantic_ai.evaluate import LLMJudge
judge = LLMJudge(model='openai:gpt-4o')
result = await judge.evaluate(question='Is Paris the capital of France?', answer='Yes')
print(result)
›Adds Agent.output_json_schema() method to retrieve the JSON schema for an agent's output type programmatically.
›Adds provider_url field to ModelResponse, used by cost() to route cost calculations correctly across providers.
›Adds prompt caching support for AWS Bedrock, reducing latency and token costs on repeated prompts.
›Allows custom clientInfo when connecting to MCP servers, enabling clients to self-identify to MCP endpoints.
›Allows model to be passed as a plain string in LLMJudge, simplifying evaluation setup.
+1 moreshow less
›Adds support for GPT-5.2 and bumps the OpenAI dependency to v2.11.0.
Pi v0.20.0 displays loaded skills on startup and adopts the SKILL.md directory convention.
└──▷ GET THIS VERSION
$ git clone --branch v0.20.0 https://github.com/earendil-works/pi.git
# already have the repo? check out this version:$ git checkout v0.20.0
›Displays loaded skills on startup in interactive mode so you can confirm which skills are active before issuing commands.
└──▷ BREAKING ON UPGRADE
!Pi skills must now be named SKILL.md inside a directory (e.g. ~/.pi/agent/skills/foo/SKILL.md); the previous convention of any *.md file directly under the skills directory is no longer supported. Migrate by renaming ~/.pi/agent/skills/foo.md to ~/.pi/agent/skills/foo/SKILL.md.
Pi v0.19.0 adds a skills system for on-demand instruction loading and a --version flag.
└──▷ GET THIS VERSION
$ git clone --branch v0.19.0 https://github.com/earendil-works/pi.git
# already have the repo? check out this version:$ git checkout v0.19.0
└──▷ TRY IT
Disable the skills system entirely when you want a clean, instruction-free agent run.
$ pi --no-skills
Check the installed Pi version quickly from the terminal.
$ pi --version
›New skills system auto-discovers and loads instruction files on demand from Claude Code, Codex CLI, and Pi-native skill directories, with descriptions surfaced in the system prompt.
›Adds --version / -v flag to print the current Pi version and exit.
Qwen Code v0.5.0 adds a VSCode Extension and ACP usage metadata tracking alongside UI copy/paste improvements.
└──▷ GET THIS VERSION
$ git clone --branch v0.5.0-nightly.20251212.8b29dd13 https://github.com/QwenLM/qwen-code.git
# already have the repo? check out this version:$ git checkout v0.5.0-nightly.20251212.8b29dd13
›Adds a channel field to client identification headers/payloads for tracking request sources.
›New VSCode Extension implementation connects Qwen Code to the IDE as a companion.
›Adds ACP usage metadata reporting to track token/usage telemetry.
›Removes vertical borders from the input prompt UI, making it easier to copy/paste multi-line content.
Qwen Code v0.4.1-nightly adds a VSCode extension and ACP usage metadata support alongside UI copy/paste improvements.
└──▷ GET THIS VERSION
$ git clone --branch v0.4.1-nightly.20251212.58d3a9c2 https://github.com/QwenLM/qwen-code.git
# already have the repo? check out this version:$ git checkout v0.4.1-nightly.20251212.58d3a9c2
›Adds ACP usage metadata tracking via the Feat/acp usage metadata feature.
›New VSCode Extension implementation brings Qwen Code capabilities directly into the editor.
›Removes vertical borders from the input prompt UI, making it easier to copy/paste multi-line content.
Phoenix v12.25.0 adds Gemini tool call support, lazy-loaded embeddings, and a null sentinel for LDAP email config.
└──▷ GET THIS VERSION
$ git clone --branch arize-phoenix-v12.25.0 https://github.com/Arize-ai/phoenix.git
# already have the repo? check out this version:$ git checkout arize-phoenix-v12.25.0
›Supports setting PHOENIX_LDAP_ATTR_EMAIL to null as a sentinel value to control LDAP email attribute mapping.
›Supports Gemini tool calls in tracing and observability workflows.
›Lazy-loads embeddings and dimension features, reducing startup overhead for large datasets.
THE AI TOOLCHAIN